Synthesis and SAR of 2-aryl-3-aminomethylquinolines as agonists of the bile acid receptor TGR5

Article abstract of DOI:10.1016/j.bmcl.2010.08.014

Optimization of a screening hit from uHTS led to the discovery of TGR5 agonist 32, which was shown to have activity in a rodent model for diabetes.

Full text of DOI:10.1016/j.bmcl.2010.08.014

Bioorganic & Medicinal Chemistry Letters 20 (2010) 5718–5721
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Synthesis and SAR of 2-aryl-3-aminomethylquinolines as agonists
of the bile acid receptor TGR5
Mark R. Herbert a, Dana L. Siegel a, Lena Staszewski b, Charmagne Cayanan b, Urmi Banerjee c,
Sangeeta Dhamija c, Jennifer Anderson c, Amy Fan d, Li Wang d, Peter Rix d, Andrew K. Shiau b,
Tadimeti S. Rao d, Stewart A. Noble a, Richard A. Heyman b, Eric Bischoff c, Mausumee Guha c,
Ayman Kabakibi b, Anthony B. Pinkerton a,
*
a Department of Chemistry, Kalypsys, Inc.,10420 Wateridge Circle, San Diego, CA 92121, USA
b Department of Biology, Kalypsys, Inc.,10420 Wateridge Circle, San Diego, CA 92121, USA
c Department of Pharmacology, Kalypsys, Inc.,10420 Wateridge Circle, San Diego, CA 92121, USA
d Department of Drug Metabolism and Pharmacokinetics, Kalypsys, Inc.,10420 Wateridge Circle, San Diego, CA 92121, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Optimization of a screening hit from uHTS led to the discovery of TGR5 agonist 32, which was shown to
have activity in a rodent model for diabetes.
Received 21 June 2010
Revised 31 July 2010
Accepted 3 August 2010
Available online 10 August 2010
Ó 2010 Elsevier Ltd. All rights reserved.
Keywords:
TGR5
GPCRs
Diabetes
Bile acids
TGR5 (also known as GPBAR1 or M-BAR)1 is a Gs-coupled GPCR
mainly expressed in the GI tract, gall bladder, spleen, lung, and pla-
centa.2 It is responsive to stimulation by bile acids, for example
cholic acid (CA, 1).3,4 Ligand binding activates adenylyl cyclase
which leads to the elevation of intracellular cAMP and subsequent
activation of mitogen-activated protein kinase pathways. Activa-
tion of TGR5 by bile acids has been shown to regulate a number of
metabolic processes, including hormonal control of energy expendi-
ture.5,6 For example, mice fed a high fat diet containing 0.5% CA
gained less weight than control mice on a high fat diet alone. Consis-
tent with this finding, TGR5 knockout mice show significant fat
accumulation and weight gain when given a high fat diet.7,8 In addi-
tion, it has also been reported that stimulation of STC-1 cells (mur-
ine enteroendocrine cells) with bile acids increased glucagon like
peptide-1 (GLP-1) secretion.9 GLP-1 improves glucose homeostasis
by several mechanisms including stimulating pancreatic insulin
secretion and inhibiting glucagon secretion. Thus, a small molecule
TGR5 agonist may be beneficial for the treatment of type 2 diabetes
and obesity. Due to the activity of bile acids on a number of recep-
tors,10 there is a need for a selective TGR5 agonist to determine
the pharmacological consequences of TGR5 activation in vivo. One
approach has been to modify bile acids to yield more selective com-
pounds, of which compound 2 (Fig. 1) is a representative exam-
ple.11–14 Our approach, described herein, has been to develop a
synthetic TGR5 agonist.15
Functional ultra-high throughput screening of the Kalypsys
compound file employing HEK-293 cells stably expressing human
TGR5 (hTGR5) and measuring intracellular cAMP as the primary
endpoint resulted in the identification of several series of TGR5
agonists, including a set of quinolines represented by 3 (Fig. 2).16
Quinoline 3 displayed an EC50 of ꢀ10
l
l
M on the human receptor
M) on the mouse receptor
but was significantly less potent (ꢁ10
(mTGR5), perhaps as a consequence of the relatively weak se-
* Corresponding author at present address: Sanford-Burnham Medical Research
Institute, 10901 North Torrey Pines Road, La Jolla, CA 92037, USA. Tel.: +1
8585251796.
E-mail address: apinkerton@sanfordburnham.org (A.B. Pinkerton).
Figure 1. Cholic acid (CA, 1) and a TGR5 selective bile acid 2.
0960-894X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2010.08.014

M. R. Herbert et al. / Bioorg. Med. Chem. Lett. 20 (2010) 5718–5721
5719
plings to give quinoline-aldehydes 7, which were then subjected to
standard reductive amination conditions with a variety of amines
to afford the desired TGR5 agonists 8. In general, high yields were
obtained for all steps.
Hydroxyl analogs (10) were accessed by demethylation of the
corresponding methoxy quinolines 9 using boron tribromide in
good yields (Scheme 2).
Figure 2. Screening hit.
quence homology (83%) between the two species.17 Given our de-
sire to assess the pharmacology of TGR5 compounds in mouse
models the identification of compounds with improved mTGR5
activity became our initial objective.
A straightforward synthesis of 3 that allowed rapid access to
analogs was developed (Scheme 1).18 Thus, substituted anilines 4
were acetylated using acetic anhydride to give acetamides 5 which
were then reacted with DMF and POCl3 under Vilsmeier–Haack
conditions to afford chloroquinolines 6. Addition of various aryl
groups were then incorporated via either Suzuki or Stille type cou-
Our SAR studies initially focused on the pendant amine of 3 (R4).
We prepared a library using a broad range of amines; some of the
key compounds described in Table 1. In general, ortho- and meta-
substituted benzyl amines were inactive (data not shown). How-
ever, we discovered that para-substitution both greatly improved
hTGR5 activity but more importantly also improved mTGR5
activity.
In particular 4-CH3 (11) and 4-Cl (12) substitution gave potent
compounds (0.88
lM and 0.58 lM on the human and 3.9 lM and
4.3 M on the mouse receptors, respectively, more than an order
l
Table 1
SAR around the amine and aryl groups
R4
R3
hTGR5 EC50
(lM)
mTGR5 EC50 (lM)
a
a
Compd
1
2
3
11
12
13
—
—
—
—
6
5.3
0.14
11.0
0.88
1.1
0.25
>10
3.9
8.5
4.3
–CH2Ph
3-Thiophenyl
3-Thiophenyl
3-Thiophenyl
3-Thiophenyl
–CH2C6H4(4-CH3)
–CH2C6H4(4-OCH3)
–CH2C6H4(4-Cl)
0.58
14
15
16
17
18
19
20
3-Thiophenyl
3-Thiophenyl
3-Thiophenyl
2-Thiophenyl
3-Furanyl
0.097
0.065
2.8
3.7
3.2
13
0.58
0.71
0.50
0.49
4.6
6.1
4.0
8.1
2-Furanyl
Phenyl
21
22
23
24
25
4-Pyridyl
0.044
0.065
>10
3.7
3-Pyridyl
5.0
O
>10
>10
>10
5.0
2.7
a
Values are means of three experiments, standard deviations were within 20% of reported values.

5720
M. R. Herbert et al. / Bioorg. Med. Chem. Lett. 20 (2010) 5718–5721
Scheme 1. Reagents and conditions: (a) Ac2O, DMAP, CH2Cl2, 70–90%. (b) POCl3, DMF, 90 °C 30–85%. (c) R3B(OH)2, Pd(PPh3)2Cl2, 2 M Na2CO3, DME, 90 °C or R3Sn(nBu)3, DMF,
90 °C, 40–70%. (d) R4NH2, NaHB(OAc)3, AcOH, iPrOH, 75–95%.
Scheme 2. Reagents and conditions: (a) BBr3, CH2Cl2, 80–90%.
Table 2
SAR on the quinoline ring
Figure 3. Compound 32 evokes increased GLP-1 release in DIO mice following
dextrose bolus. *p <0.05 for vehicle.
R1
R2
hTGR5 EC50
(l
M)
mTGR5 EC50 (lM)
a
a
Compd
15
26
27
28
29
30
31
32
33
34
35
36
37
38
–OCH3
–CH3
–Cl
–OEt
–OCF3
–CO2H
–H
–OH
–H
–H
–H
–H
–H
–H
–H
–H
–H
–H
–OCH3
–OH
–OH
–OCH3
–OCH3
–OH
0.065
5.0
6.9
3.2
>10
>10
9.4
>10
>10
>10
0.28
>10
1.6
0.32
>10
>10
>10
stitution led to a large loss in potency. A phenyl group (20) was tol-
erated, showing comparable potency to the furan and thiophene
4.5
>10
>10
5.5
substituted compounds (hTGR5 EC50 of 0.49
lM, mTGR5 of
8.1 M). Pyridyl substitution showed activity comparable to 15,
l
with 4-pyridyl compound 21 (hTGR5 EC50 of 44 nM) and 3-pyridyl
compound 22 (hTGR5 EC50 of 65 nM) both being among the most
active human compounds tested. Lastly, a compound with a het-
eroatom linker, as in hydroxypyridine 23, showed no activity.
We next turned our attention to the SAR around the quinoline
ring, using 15 as a starting point (Table 2). In general, substitution
at the 4, 5, 6, and 8 positions of the quinoline were not tolerated
(data not shown), with the exceptions at the 6 position noted be-
low. Examining a range of substituents at the 7 position, we ini-
tially found that only methoxy was well tolerated. Methyl (26),
chloro (27), ethoxy (28), trifluoromethoxy (29), carboxylic acid
(30), and hydrogen (31) all gave only weakly potent or inactive
compounds. Interestingly, demethylation of 15 to give phenol 32
gave a compound that was considerably more active on the mouse
5.1
>10
0.082
0.43
>10
>10
>10
–OCH3
–OCH3
–OH
–OH
a
Values are means of three experiments, standard deviations were within 20% of
reported values.
of magnitude improvement over the initial lead and comparable to
the potency of cholic acid). Extension of the benzyl amines to
phenethylamines such as 14 and 15 led to a further enhancement
in potency, especially on the human receptor. Compound 15, which
receptor (0.28
lM vs 3.2
lM) but less active on the human recep-
displayed an EC50 of 65 nM on the human receptor and 3.2 lM on
tor (5.1 M vs 0.065
l
lM). Due to our interest in developing a
the mouse receptor, became our point for further optimization. We
next focused on the 3-thiophenyl ring (R3) with the results sum-
marized in Table 1. Other five-membered heterocyclic rings, such
as the 2-thiophenyl (17), 3-furanyl (18), and 2-furanyl (19) were
somewhat less potent than 15. Thiazole (24) and oxazole (25) sub-
mouse active compound for in vivo studies, we investigated this ef-
fect further. As with other groups, methoxy substitution at the 6
position (33) gave an inactive compound. Intriguingly, demethyla-
tion to 6-substituted phenol 34 gave a compound that was again
Table 3
Selected mouse PK parameters for compounds 15, 26, and 37a
AUCc
(lg h/mL)
t1/2 (h)
%F
b
b
c
c
Compd
Clp (mL/min/kg)
Vd (L/kg)
Cmax
(
lg/mL)
32
35
38
29
2.4
1.4
0.37
0.50
0.40
0.86
4.4
7.7
9
10
a
b
c
Compounds dosed 3 mpk iv and 30 mpk po.
iv.
po.

M. R. Herbert et al. / Bioorg. Med. Chem. Lett. 20 (2010) 5718–5721
5721
3. Kawamata, Y.; Fujii, R.; Hosoya, M.; Harada, M.; Yoshida, H.; Miwa, M.;
Fukusumi, S.; Habata, Y.; Itoh, T.; Shintani, Y.; Hinuma, S.; Fujisawa, Y.; Fujino,
M. J. Biol. Chem. 2003, 278, 9435.
4. Houten, S. M.; Watanabe, M.; Auwerx, J. EMBO J. 2006, 25, 1419.
5. Watanabe, M.; Houten, S. M.; Mataki, C.; Christoffolete, M. A.; Kim, B. W.; Sato,
H.; Messaddeq, N.; Harney, J. W.; Ezaki, O.; Kodama, T.; Schoonjans, K.; Bianco,
A. C.; Auwerx, J. Nature 2006, 439, 484.
6. Groen, A. J. J. Hepatol. 2006, 45, 337.
7. Maruyama, T.; Tanaka, K.; Suzuki, J.; Miyoshi, H.; Harada, N.; Nakamura, T.;
Miyamoto, Y.; Kanatani, A.; Tamai, Y. J. Endocrinol. 2006, 191, 197.
8. Vassileva, G.; Golovko, A.; Markowitz, L.; Abbondanzo, S. J.; Zeng, M.; Yang, S.;
Hoos, L.; Tetzloff, G.; Levitan, D.; Murgolo, N. J.; Keane, K.; Davis, H. R., Jr.;
Hedrick, J.; Gustafson, E. L. Biochem. J. 2006, 398, 423.
9. Katasuma, S.; Hirasawa, A.; Tsujimoto, G. Biochem. Biophys. Res. Commun. 2005,
329, 386.
10. Makishima, M.; Okamoto, A. Y.; Repa, J. J.; Tu, H.; Learned, R. M.; Luk, A.; Hull,
M. V.; Lustig, K. D.; Mangelsdorf, D. J.; Shanz, B. Science 1999, 284, 1362.
11. Pellicciari, R.; Sato, H.; Gioiello, A.; Costantino, G.; Macchiarulo, A.; Sadeghpour,
B. M.; Giorgi, G.; Schoonjans, K.; Auwerx, J. J. Med. Chem. 2007, 50, 4265.
12. Sato, H.; Macchiarulo, A.; Thomas, C.; Gioiello, A.; Une, M.; Hofmann, A. F.;
Saladin, R.; Schoonjans, K.; Pellicciari, R.; Auwerx, J. J. Med. Chem. 2008, 51,
1831.
13. Sato, H.; Genet, C.; Strehle, A.; Thomas, C.; Lobstain, A.; Wagner, A.;
Mioskowski, C.; Auwerx, J.; Saladin, R. Biochem. Biophys. Res. Commun. 2007,
362, 793.
Figure 4. Compound 32 promotes increased glucose clearance in DIO mice during
OGTT. *p <0.01 for vehicle.
14. Pellicciari, R.; Gioiello, A.; Macchiarulo, A.; Thomas, C.; Rosatelli, E.; Natalini, B.;
Sardella, R.; Pruzanski, M.; Roda, A.; Pastorini, E.; Schoonjans, K.; Auwerx, J. J.
Med. Chem. 2009, 52, 7958.
active on both the human (0.082
(1.6 M). Combining these results led us to compound 35, which
displayed potency both on the human (0.43 M) and mouse
(0.32 M) receptor. Other combinations of methoxy and phenol
groups were not active (36–38).
Mouse pharmacokinetics studies were carried out on 32 and 35
(Table 3). In general, these compounds demonstrated moderate to
high clearance, with modest bioavailability. However, the exposure
after oral dosing of 30 mpk was sufficient to investigate the effect
of the compounds in vivo.
To test the hypothesis that activation of TGR5 stimulates GLP-1
release, diet-induced obese (DIO) mice were dosed with compound
32 prior to an oral glucose tolerance test (OGTT). As shown in Fig-
ure 3, a 30 mpk oral dose of 32 increases plasma GLP-1 following
an oral glucose challenge.19,20
We also examined the effect of 32 during an oral glucose toler-
ance test (OGTT). As shown in Figure 4, acute dosing of 32 en-
hanced glucose clearance significantly, consistent with our
previous finding of elevated plasma GLP-1 levels.21
In conclusion we have developed a series of quinolines that are
both potent and selective22 agonists of human and mouse TGR5.
Activation of TGR5 was shown to increase GLP-1 secretion and en-
hance glucose clearance in vivo, warranting a further investigation
of TGR5 agonists as potential therapeutics for the treatment of type
2 diabetes. Future reports from our group will focus on the further
optimization of these molecules as well as a more in depth discus-
sion of the pharmacology studies.
lM) and mouse receptor
l
l
15. Recent reports of other synthetic or natural product derived TGR5 agonists can
be found in: (a) Evans, K. A.; Budzik, B. W.; Ross, S. A.; Wisnoski, D. D.; Jin, J.;
Rivero, R. A.; Vimal, M.; Szewczyk, G. R.; Jayawickreme, C.; Moncol, D. L.;
Rimele, T. J.; Armour, S. L.; Weaver, S. P.; Griffin, R. J.; Tadepalli, S. M.; Jeune, M.
R.; Shearer, T. W.; Chen, Z. B.; Chen, L.; Anderson, D. L.; Becherer, J. D.; De Los
Frailes, M.; Colilla, F. J. J. Med. Chem. 2009, 52, 7962; (b) Genet, C.; Strehle, A.;
Schmidt, C.; Boudjelal, G.; Lobstein, A.; Schoonjans, K.; Souchet, M.; Auwerx, J.;
Saladin, R.; Wagner, A. J. Med. Chem. 2010, 53, 178.
16. HEK293 cells stably expressing human or mouse TGR5 were established by
stably transfecting HEK-293 cells with an expression vector (pcDNA 3.1,
Invitrogen) inserted with human TGR5 cDNA using Fugene6 (Roche,
Indianapolis, IN) according to conventional methods. Cells were grown in
DMEM (Invitrogen, Carlsbad, CA) supplemented with 10% FBS, 1% penicillin/
streptomycin under geneticin selection. To assess the activity of test
compounds, cells were harvested using non-enzymatic cell dissociation
buffer (Invitrogen, Carlsbad, CA), seeded in DMEM supplemented with 0.1%
FBS at 0.8 Â 106/mL and incubated overnight at 37 °C in an atmosphere of 10%
CO2 and 95% humidity. The next day, test agents were added to cells in the
presence of 1 mM IBMX and incubated for 30 minutes at 37 °C. Intracellular
cAMP was then measured by TR-FRET using a commercially available LANCE
kit (Perkin–Elmer, Boston MA). All compounds were tested in 20-point dose
response, n = 3. Compounds were full agonists on both the hTGR5 and mTGR5
receptors, comparable to cholic acid.
l
17. Thomas, C.; Pellicciari, R.; Pruzanski, M.; Auwerx, J.; Schoonjans, K. Nat. Rev.
Drug Disc. 2008, 7, 678.
18. Complete experimental procedures and spectral data can be found in:
Pinkerton, A. B.; Kabakibi, A.; Herbert, M. R.; Siegel, D. L. WO2008/097976.
19. Preliminary data presented at: Guha, M.; Pinkerton, A.; Banerjee, U.; Dhamija,
S.; Anderson, J.; Herbert, M.; Siegel, D.; Hoffman, T.; Staszewski, L.; Cayanan, C.;
Shiau, A.; Rao, T.; Noble, S.; Kabakibi, A. National Meeting of the American
Diabetes Association, San Francisco, CA, June, 2008.
20. Compound 32 as a bis-HCl salt (30 mpk) in CMC (0.5%)/Tween 20 (0.25%) was
dosed orally to DIO mice (n = 6 mice per time point, fasted for 18 h) 15 min
prior to an oral dextrose bolus (2 g/kg in saline). Blood was collected from the
tail vein at 3, 6, 9, 12, 15, and 20 min post dose. To prevent the degradation of
GLP-1, a sub-optimal dose of vildagliptin (0.1 mpk) was given 25 min prior to
dextrose dosing. Plasma GLP-1 was measured by ELISA. Data is represented as
plasma GLP-1 level at 9 min post dextrose dose.
21. Compound 32 bis-HCl salt (30 mpk) in CMC (0.5%)/Tween 20 (0.25%) was
dosed orally to DIO mice (n = 6 mice per time point, fasted for 18 h) 15 min
prior to an OGTT (2 g/kg dextrose in saline). Blood was collected from the tail
vein at 0, 15, 20, 30, 60, and 120 min post OGTT, and plasma glucose level was
measured using a hand-held glucose meter. Data is represented as glucose AUC
post OGTT.
Acknowledgments
We thank Robyn Rourick and Nahid Yazdani for determining
plasma levels for PK parameters and Nicholas D. Smith for helpful
comments.
References and notes
22. Compounds 32 and 35 were screened in a CEREP panel of approximately 30
targets that could potentially affect glucose or GLP-1 levels and their activity
1. Tiwari, A.; Maiti, P. Drug Disc. Today 2009, 14, 523.
2. Maruyama, T.; Miyamoto, Y.; Nakamura, T.; Tamai, Y.; Okada, H.; Sugiyama, E.;
Nakamura, T.; Itadani, H.; Tanaka, K. Biochem. Biophys. Res. Commun. 2002, 298,
714.
was found to be <50% of optimal at 10
action.
lM confirming a TGR5 mechanism of